IscoRuta98/thesis-code

Code repository for masters thesis.

Detect.py

• A python script that uses pretrained Deep Learning models to detect the number of: figures, equations,and tables in a particular article.

Assumptions

• Python is installed on your local machine.
• You have the following in your working directory:
  • Detect.py script
  • requirements.txt script
  • output.csv: Keeps records of the equations, figures, and tables for a particular article.
  • Master-list of a particular journal can be downloaded here.
  • Pivot list of a particular journal can be downloaded here.
  • Articles pertaining to a particular journal (You can download them here).
• You are either using Linux OS or macOS system. Dependencies such as Deetectron2 (deep learning model used to detect tables and figures) are quite tricky to install on Windows operating systems
  • Click here for additional Instruction to installing Detectron2 Layout Model for Windows users.

Installation & Running Detect.py script

• In your working directory run the following command to install the python packages required for Detect.py script:
  • pip install requirements.txt
• In the Detect.py script, replace the following variables with the directory the files (i.e., master list, pivot, output Excel files, and directory to the articles on your local machine) are in:

masters = pd.read_excel("/working/directory/master.xlsx")
pivots = pd.read_excel("/working/directory/pivots.xlsx")
output = pd.read_csv("/working/directory/output.csv")
directory = r'/directory/to/stored/papers/

• To run the Detect.py script, run the following command in your terminal:
  • python Detect.py